Paloma Piot

I am a postdoctoral researcher in Natural Language Processing at the IRLab, University of A Coruña, Spain, where I recently completed my PhD (Cum laude, 2026) as a Marie Skłodowska-Curie Actions (MSCA) doctoral candidate within the HYBRIDS project. My research focuses on how large language models (LLMs) perceive, assess, and sometimes reproduce hate speech and harmful content in social media. I am particularly interested in the intersection of AI and society, examining how language technologies can reflect social biases and how they can be developed to be more responsible, transparent, and trustworthy.

My work centers on hate speech and abusive language detection, multilingual and low-resource NLP, and the evaluation of LLMs across different languages, cultures, and communities. I investigate the social and sociolinguistic dimensions of online discourse, with an emphasis on fairness, explainability, and robustness in language technologies. Looking ahead, I am keen to explore a perspectivist direction: hate speech is inherently subjective, and I want to understand how LLMs handle the differing judgements of the groups most affected, rather than collapsing them into a single label. This builds on, my work on bias, efficiency and explainability. I am also interested in data-centric approaches, synthetic data generation, and the development of datasets and evaluation frameworks that help improve the detection and mitigation of online harm.

I am committed to open and reproducible research, and I maintain open-source implementations of my work and related projects on GitHub. You can also follow my academic and professional updates on LinkedIn and X.

Feel free to reach out via email at <paloma.piot[at]udc.es>.